“…As crowdsourcing has become an increasingly recognized means of data collection (Eskénazi, Levow, Meng, Parent, & Suendermann, 2013), researchers have recruited AMT workers to evaluate mispronunciations in native (McAllister Byun, Halpin, & Szeredi, 2015) and L2 speech (Peabody, 2011;Wang, Qian, & Meng, 2013), and to rate samples for features such as accentedness (Kunath & Weinberger, 2010). In addition to providing access to a larger and possibly more representative sample of listeners across a range of languages, services like AMT may prove to be a necessary data collection tool for designs that generate a large number of samples, such as the longitudinal pronunciation studies that are becoming more common in L2 speech research (e.g., Derwing & Munro, 2013;Nagle, 2018b;Saito, Dewaele, Abe, & In'nami, 2018).…”